An Analytical Model of Light Scattering from Marine Micro-Organisms and Detritus.
Abstract
The design and initial training of an artificial neural network that predicts the size parameter of a light scatterer given its S34 matrix element was completed. As an alternative to experimental data, the network was trained and tested for both spherical and irregularly-shaped particles using Mie calculations and calculations made with the coupled-dipole model. The important features of a network are illustrated with this simple version. A more extensive network that predicts optical properties and shape factors, as well as size parameter, has also been designed. It does not differ in basic features from the simpler one, although it requires more input data and many more neurons to produce the desired results. We have shown that Mie and coupled-dipole calculations can provide a workable data set for training a neural network. However, it is important to test the network with experimental data.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 03, 1995
- Accession Number
- ADA308259
Entities
People
- Patricia G. Hull
Organizations
- Tennessee State University